A prospective multicenter clinical research study validating the effectiveness and safety of a chest X-ray-based pulmonary tuberculosis screening software JF CXR-1 built on a convolutional neural network algorithm

Background Chest radiography (chest X-ray or CXR) plays an important role in the early detection of active pulmonary tuberculosis (TB). In areas with a high TB burden that require urgent screening, there is often shortage radiologists available to interpret results. Computer-aided (CAD) software employed artificial intelligence (AI) systems may have potential solve this problem. Objective We validated effectiveness and safety imaging screening based on convolutional neural network algorithm. Methods conducted prospective multicenter clinical research validate performance (JF CXR-1). Volunteers under age 15 years, both without suspicion tuberculosis, were recruited for CXR photography. The reported probability score each participant. results compared those by radiologists. measured sensitivity, specificity, consistency rate, area receiver operating characteristic curves (AUC) diagnosis tuberculosis. Besides, adverse events (AE) severe (SAE) also evaluated. Results was six general infectious disease hospitals across China. A total 1,165 participants enrolled, 1,161 enrolled full analysis set (FAS). Men accounted 60.0% (697/1,161). Compared from board, showed sensitivity 94.2% (95% CI: 92.0–95.8%) specificity 91.2% 88.5–93.2%). rate 92.7% (91.1–94.1%), Kappa value 0.854 ( P = 0.000). AUC 0.98. (SS), which consisted participants, 0.3% (3/1,161) had AEs not related software, no observed. Conclusion algorithm effective safe. It candidate solving problems lacking burden.